How AI is Redefining the Finance and Headcount Playbook
Podcast Overview
-
Eric Guidice: Headcount experts, episode 17. Chris, Manny, and Eric Guidice. We have a special guest, Reed Gilbert, joining us. Welcome to the show, welcome to the podcast. We're very excited to have you on. How do you and Chris know each other?
Reed Gilbert: Gosh. We go back to Wayfair. Which is a long time ago.
Chris Mannion: Yeah, so Reed and I worked together at Wayfair for a few years and I worked pretty closely with him when I moved into the talent space and then kind of watched his career as he moved into senior finance leadership roles, various different companies. We stayed in touch since then. He's taken on HR ownership as well in some of those roles. And so I think has a very interesting perspective of the overlap between finance and HR. And whereas I think some of us that maybe will watch this, have that frustration going into a QBR when they're trying to get their message across and don't really understand what finance are asking. I think hopefully we're gonna clarify some of those things today. So Reed, I think that kind of covers anything else to add.
Reed Gilbert: You've said it in a much nicer way than I would. I took a series of increasingly broad roles at increasingly smaller companies. And so as part of that, I had to phone in help from all the people I'd worked with across the different teams at Wayfair and other companies. And I was lucky to be able to lean on Chris for advice on the recruiting space.
Eric Guidice: I love it. Okay, well I think one of the main things that we hit on in this conversation or, you know, for the different guests that we have on is this is like one of the few places where all these different teams come together with some sort of common denominator. And each individual team has their perspective or a bias sometimes or like their own goal to hit and the relationship between the different teams that come together over this data set is just something that we've found a tremendous amount of interest in exploring not only because we have our own businesses or we're providing services in the space but because it's just a genuinely interesting spot for a business to kind of come together and explore. So I'm curious to know kind of like you know in the recruiting space to tell me about but like give us the history of your career in this experience. Where did the scope start? Where are you ending today and like, what perspective are you bringing to the headcount conversation?
Reed Gilbert: Yeah, so the really quick walk is I started off as an investment banking analyst. And then after business school moved into corporate finance role at Wayfair and then into operations roles and then leadership at series A startups. So I am in my third startup where I'm wearing the finance and CFO type hat. And headcount is one of the things that matters most from a cost side, from a people perspective. Do you feel excited to come to work because you're with the right people. And from a success side, if you don't have the right people, you're not set up for success. It's almost always the longest pole in the tent as far as the thing you have to get right before any of the other pieces can fall into place. And so today I'm at Reprise. We're an AI demo company where you can use our MCP and our tools to customize a demo for whoever you're selling to using all of your transcripts and information that you have on hand. So you give them the targeted pitch leveraging all the info you have and you deliver a demo that doesn't crash and burn.
Eric Guidice: These are async demos. So, you know, it had for my particular business headcount 365. Sometimes we'll sell to a recruiting person. Sometimes we'll sell to a finance person, sometimes to a budget owner or like a seller who's like, how many salespeople do I need? And so we use Vidyard today to like give out, hey, here's the demo for you. But what you're saying is, is they can interact with an AI that would allow us to like customize a demo in a highly tailored fashion to that person async. Am I getting that right?
Reed Gilbert: These are leave behind demos that are interacted with guys, and live demos where instead of taking somebody through a tenant on your production app, you can take somebody through a front end replicate copy where it looks and it seems exactly like the real thing, but it's in a safe environment where it's not calling the backend. So you eliminate the chance that a bug an engineer put into production 15 minutes ago, takes down your demo and your sales prospect.
Eric Guidice: Got it. How many people are Reprise today?
Reed Gilbert: So that's a little bit of a story. Today we are 20 people and the impact of AI on what our team has been able to do has been astounding. It's honestly hard to start talking about headcount and team planning without talking about AI first because they're two sides of the same coin.
Eric Guidice: And Chris, I want I want to I posted a clip just the other day of us doing a previous podcast of like how the AI narrative is shaping like the value prop to CFOs. And then I was on LinkedIn yesterday. I'm talking to Chris Abbess from Talentful. He went to the Bright Higher Shine conference. He's saying, he believes for finance folks, there will be like a token budget with headcount as well. And so I'm curious to get from you kind of the where you think it's going from a, so if the conversation about headcount can't be had without talking about AI, well, what are you seeing? What are the specifics about headcount and AI that are coming into the conversation from your lens?
Reed Gilbert: The biggest thing for us is the rate of change with AI. If you look at what Cowork and Claude and OpenAI are releasing week in and week out, you don't really know where the ball's gonna be in a month's time or in two months time. And your headcount decisions have a 12 month horizon because it takes you time to fill the role, time to ramp somebody, and then hopefully they achieve the intended result.
Chris Mannion: It sounds like the barrier to getting this right is so much higher now and the ability to forecast exactly what you're going to need is so much bigger. There's this very interesting analysis you need to do where you're not kind of hiring a hundred people and if 10 don't work out, it's not the end of the world. If you're only hiring 10 people, you need every one of those to be the right fit. How does that change in the planning process and then the hiring, the design of the hiring structure?
Reed Gilbert: We're lucky that being such a small team, we keep structure really low. And the number of roles I'm recruiting for right now is one. We are looking for a jack of all trades marketer. Somebody that is embracing AI because it's too soon to say what this person will need to do in the future. We just need somebody who is excited to adapt and lead that charge for us. So when it comes to forecasting, I always try to be faster and more responsive than more perfect in my predictions of the future. So if that's decreasing the time to hire, decreasing the ramped time. Decreasing the decision making cycles and just say like, okay, we roughly have the profile we're going for, but until we meet the candidates, we're not gonna know exactly who, let's just get started. It has us in this sort of stop and go motion where we pause, we take it as long as we can before we start doing something. And then we do something with extreme urgency, even compared to what we were doing six months ago.
Eric Guidice: So I saw a very interesting post from the head of talent at Calendly, Logan Marsh. He said, and it was about the block hiring where he said the org chart will change from a pyramid to a circle. That was the, you know, if I had to summarize the whole article and then his thoughts about it, where instead of it being hierarchy of humans, there's someone at, the AI is at the center and then there's humans around the outside kind of interacting with this space. So it sounds like from a marketing perspective, that's, you know, you might be leaning on existing models and then that person's ability that you've already purchased, right, tokens or models that you've already purchased, and then that person's ability to interact with that to instead of being, you know, a demand gen and a content marketer or XYZ, you're just saying one person, you do all of it, and these agents are now your employees. That's what I'm hearing.
Reed Gilbert: Yes, there is a lot of that, although that's a little bit future state because we've got to build it first. And this particular hire has to lay most of that groundwork, but it's already changing how we think about longevity of employees versus longevity of your AI information. And some of the AI tools that we're building, we expect to persist forever.
Eric Guidice: Well, let me ask you this question, because it's very interesting that you say that. The structure of the AI information within a company being something that can outlast an employee who may take a promotion or have a situation they need to leave work for, etc. Are you now creating a new job or do you envision creating a new job around the management or maintenance of this AI information infrastructure? Is that a whole other thing that you're looking at or someone owned that today? What's your take on people versus that data infrastructure?
Reed Gilbert: So today we have a huge emphasis on AI fluency and adoption and we're insisting that everyone do it for themselves. Like you can't move away from this fire and so hiring a role to own that would sort of take ownership away from other people. At a larger organization where there's more cross-functional coordination or ownership needed, I could see that, but for a series B startup like us, this is just something we all have to contribute to and keep top of mind as we build our tools and automate the processes.
Eric Guidice: Okay, how does it change your... Oh, go ahead, Chris.
Chris Mannion: I think there's an interesting angle there. We did a few episodes back, we did a focus on capacity planning, because that's something that a lot of talent leaders spend a lot of time doing and trying to figure out what is the capacity needed for the organization. And then back into what is the capacity I need on my recruiting team in order to deliver that organizational capacity. I'm actually curious how you think about that now, if you think about the workforce and you almost have these two layers where you have the people element and then the productivity of the AI element. And it sounds like that's layered on top of each individual. It's not, there's no separate entity that is, you know, this is the AI efficiency team. Like everyone is responsible for that. So I'm curious how you think about that in terms of capacity expectations for the individuals. How have you seen that change over time? Then is that, do you think we've reached a point where that is relatively static or do you foresee that kind of continue into change and how does that back into capacity planning for the overall organization?
Reed Gilbert: We've thrown out all the plans that we did three months or four months ago and we're just not planning. It's like every week AI will do something for me that I used to do myself and it would be a chore or a task. Like yesterday it did my quarterly board meeting option pool roll forward summary. Just give it the cap table download all of the transaction level detail. Tell me how many options are left. Tell me which ones were returned and came back. Which options are we about to grant? Two minutes, like here's the slot. And so like that's just an example where we're all seeing new things like that every day. And we just don't know when it's gonna stop. And we know that we don't understand the magnitude of it. So we're just all executing. And when we hit the wall we can't run through, that's when we'll talk about, do we really need a wreck for this? And are we sure? And how do we know that we won't be wrong in two weeks time when the next model drops?
Eric Guidice: Are you budgeting additional costs in a person's like fully burdened headcount cost for these new models. So where it used to be salary benefits, software costs, amortized office equipment, et cetera. Is this a separate line item or are you just rolling it into a different cost model or do you not care about it at all because of the productivity from this process so outweighs the cost that it's kind of just like, okay, that's fine. Because I've seen some articles where in more of the engineering space, they're like, we're hiring people because it helps us save money on AI. Is that something that you're factoring into the financial modeling?
Reed Gilbert: So I'll tell you how we're doing it and, I had the funniest exchange as a leadership team because every once in a while, Claude will block our access because we've exceeded our usage. And it's like a five alarm fire of how do we throw money at this vendor so we can spend more, which is something that's never the case with other software or technology providers. Like it's a good example of why it's a category of its own. And the way we're budgeting and planning for it as an organization is actually not headcount based because I kind of assume that no matter how many people we have, our token consumption is going to increase month over month, which is just the way we look at budgets. And so I've got a growth rate factor then. We'll see if that holds true or not. We're starting to be a little more, really, not sophisticated, but rudimentary about which models we use for which purposes to try and constrain our spend versus cleaning up after the fact. But this is the type of spend, just like you said, Eric, that we're very happy to take because the benefits are super clear.
Eric Guidice: Sorry, so if I hear you correctly, like the two things that might influence token usage are the amount of people using it on the headcount side, but that might not require its own individual person modeling. But it's also the capability of the model. So the more things that it could do, the more likely you are to use it. So if there's additional functionality or a model that comes out that might be able to solve another problem, it's much easier to allocate spend there because it's an immediate ROI versus higher ramp and then plan an individual person to do it. So you might expand a capability in the model and then might graduate to a person who then runs that model among other things. But those are the two kind of inputs to whether or not you might expand spend on AI.
Reed Gilbert: Yes, although one of the interesting things is I think our spend is going to increase independent of our headcount. Like, I expect the spend per person to rise. I expect that spend to replace spend that would have gone to headcount. And so that's another reason why we're happy to just see how far, how productive we can make each employee. You have all this overhead cost of orchestrating a team and keeping people aligned and in sync. And that's cost that goes down when you don't make hires. So there's a bunch of spend that's harder to put your finger on that we're seeing the benefit of, like our headcount has shrunk over the past six months and it's meant that we've gone faster. It means that our company plans and sets goals on shorter time horizons because we're moving more quickly. And we're just, we're so convinced that this trend will continue that it's like really exciting to come to work every day because you don't know how it's going to surprise you next. And there's something too, just about startups and building things and working with AI and building things. And it's like the people who like to be at small companies because they can feel the results and they have all this exposure to the features that you're shipping. There's a personal gratification that you see from using your command bar to get work done with an agent.
Chris Mannion: Yeah, follow up question on the productivity. I've spent the last couple of months just speaking to dozens of people in the kind of leadership layer, rolling out AI tools. And one of the biggest takeaways that I got from that was that the cost of the tools is so small that many are just registering all employees with a license for whatever tools are available and not really setting any productivity benchmarks or goals for the teams. It sounds like at a small company, you actually have much more visibility and much better way of measuring that. Is that true? Did you come up with a way to measure productivity gains? So you look at 1 dollar of AI spend equals X dollars of productivity gains per person. And if so, how would you recommend people think about that if they've got like a 10,000 person organization and they're trying to figure this out at scale.
Reed Gilbert: We have seen especially in the engineering side that AI is so self-evidently productive that we actually haven't spent many iterations trying to value just how productive it is. When we look at the amount of code we've shipped, how much of that code has to get edited or revised after it's been pushed, and the proportion that's authored by AI, everything is up and to the right exponentially. And so long as that continues, we're just putting the incremental time into doing the work rather than trying to extrapolate where we think it'll be a couple quarters from now because the rate of change is so high.
Chris Mannion: It sounds like you're not even kind of breaking down the productivity gains. It's just a new expectation for the capacity of an engineer to what they can ship. That is just completely changing your model of how you structure an organization.
Reed Gilbert: Yes, it's totally changing the model because our designer is now one of our largest code contributors. The product managers are shipping their code and it's forced our engineering team to double down on the DevOps side of the house because they're in charge of the code base and the architecture quality. So they are, their job is to help everyone at the company commit and keep our stability, security, and reliability where it needs to be. And so it's not just changing the productivity of the engineering team, it's changing how all of the teams in the build side of the house work together.
Eric Guidice: I have a bit of a question that's, let's call it a curveball, but I'm just going to give you my perspective from my Uber days where during the build process, like taxis offered a ride for a price. When Uber was in the fundraising hype era, it was a much lower price for a much better quality that now that they're a public company, we've kind of seen those things come together. What price right as OpenAI considers an IPO as we start to hit the limitations and I'm sure you're obviously not the only CFO talking in this way so that is going to you're going to there you're priced into a place where now you have to use it because of all these gains that are just so great and what price do you start considering the human verse AI conversation like obviously it's a subsidized process now but at what point how much can these Netflix charge me before I'm like why am I paying 28 dollars to watch Seinfeld reruns do you know what I mean?
Reed Gilbert: Yeah, it's a question I hope I don't have to grapple with for a long time because right now we're in this zone where it's an economic win for everybody, at least on the beneficiary or the business consumer side of the table. Thinking in hypotheticals as like, I'm sure the prices will rise and every time they retokenize, we have uncertainty over what the impact of that will be. But for us, I think it would be less about when will we start adding more humans to manually augment capacity versus how few people does it really take to run this automated system.
Eric Guidice: That's super interesting. I think when there's always a conversation, so whenever I've been in a workforce planning conversation, we're looking at not only the number of people, but the performance of those people in sort of like this nine box style. Who do we want to invest in? Who do we grow? How do we spend more? How many people do we need? If we elevated people from one area of the nine box to another, what is the gains for us? Is there any sort of, are you evaluating these things on performance? And is it all objective like code shipped or is there another metric that you're considering as you look at like, okay, where do we invest? Why do we invest? How do we invest? Whether it's a headcount or an AI. Is there like that, I don't want to call it a nine box exercise, but is there something that you're doing? What are your metrics to measure at any cadence, whether it's a quarter or a year for the board deck?
Reed Gilbert: So, it's kind of funny, I'll go back to Wayfair for this. At Wayfair we had nine boxes and it was also true at Morgan Stanley where I was beforehand where you'd have the like core evaluation and then you'd get extracurricular credit for different things. Like if you were doing campus recruiting that was a positive or if you were involved in employee engagement stuff there are other things I'm forgetting but you get the flavor and six to twelve months ago like an AI tinker was a type of extracurricular where it's like there's an intangible bonus for that. Now all the AI metrics is just embodied in your core productivity. It's like what can a software engineer do and how do they compare against each other? It's like we've gone back to the basic metrics because the operating assumption is that everything is AI infused and you see stark differences that make it even easier to make talent improvement or correction activities where some people are adopting it and other people aren't.
Eric Guidice: What about from like a, let's call it compensation or merit strategy? And I understand we might be in a sensitive area here, so we can operate as such. But as you're looking at the way an individual produces for the company, how are you with AI changing so much, associating merit with productivity? And how are you factoring in the cost of an employee or salaries? Is it a, well, the supply of people is going up to the interview process because AI is reducing the amount of total jobs in the market and so therefore we could pay less or is it an individual is contributing so much more to the business we're rewarding them with more compensation through merit increases in a measure of productivity like where what's the lean here and I understand the nature of that question.
Reed Gilbert: Yeah, so like I'm thinking hard about what is fair to say versus what is too much of a leading edge hot take. Truthfully, there are sort of two basics here. One is that the way we did compensation is still the way we do compensation, where it's based on industry benchmarks and you look at what other similar firms are paying and you decide what your talent philosophy and what your comp philosophy is and where you want to be in the market. You know, market goes up, market goes down, and market stays the same. Like you try to maintain your spot. And so that hasn't changed a lot for us. One of the things that has changed is the talent strategy and who we're trying to hire. We're looking for people that want to spend tokens, that want to try stuff and we don't cap people or cut them off when they hit a certain point. We're not at that stage yet where we don't think there are further marginal returns or that we have a good enough sense for here's how many tokens or what your dollar spend is to do this job. And so it's not part of compensation, but it's part of the value proposition that if you come work at Reprise, you get unlimited access to the tools you need to explore and experiment.
Eric Guidice: Yeah, I think people take jobs for two reasons. That it's better for you today or it's better for your future. And so if you get to have unlimited experimentation with whatever model you want to create a business outcome that you get to take credit for at a company of this size, that's something I would measure in my job search. I want to get like Chris get one in. I could tell he's interested, but I also have questions on what this has done to your fundraising strategy and how the headcount AI relationship has changed. You know, the way that you think about ownership and fundraising and what dollars you get and how much do you need to dilute, et cetera. Like there's a lot of, you know, I sell a lot of headcount solutions to tactical users who are trying to do what AI is doing for your engineering team with this process of headcount. However, the argument is that there's a much bigger benefit to understanding the relationship of your headcount data to the business. So the fundraising thing is the next place I want to go. But before I go there, Chris, I could see you reacting to some of the things that Reed's talking about. What's on your mind?
Chris Mannion: Yeah, I'm just kind of thinking back to, we just did a couple of episodes on the QBR and that process. And I think your example, Reed, of how you're actually doing your kind of pool projections now in two minutes. And I can imagine a lot of that process has changed from, you know, when we used to do the SDO meetings back at Wayfair, which was, if you don't know, every six months, it would kind of go deep and each business unit would essentially present out. I would imagine that things are very different now actually. Not just working at a small company, but also the expectation of having clean data and turn around very quickly and the time lag in that data. But specifically from a HR lens, I think we know that within the HR space, we're probably a little further behind on the adoption of AI tools and actually using them in the workflow. So I'm actually curious what you've seen there. I think before, I think Eric's got some really good follow-on questions, but just in terms of making practical recommendations for the people that usually watch this show, how they can actually better partner with their finance peers, whether that's an early stage where things are turning very quickly or even at much larger companies, where I think the expectations have still changed in terms of the ability to deliver cleaner data and speak to the data more accurately. What are you seeing as the changes since our time at Wayfair six years ago building out the SDO decks? And then what do you think's missing if you could ask, you know, a whole swath of HR folks to invest in certain things, like what would you ask them to do?
Reed Gilbert: Well, look, if somebody could build an HRIS system that I actually enjoy using, that's like, I think that's still the holy grail. One of the things I'm really focused on is avoiding tools and systems and using AI to scale without needing to add planning tools. At 20 people, we are so far beneath the cut point. But when I was at HQO before this, between 100 and 200 people, it was really hairy and a real pain in the butt. And these types of challenges that we were struggling through with spreadsheets, Claude in Excel is a total game changer.
Eric Guidice: Yeah, I think one of the issues with at least from my perspective, right? One of the issues with HR systems for the finance audience is the nature of the data and the integration back to a finance either spreadsheet or tool. So there's a what has happened in an HRIS, but there's this also there's a stuff that's caught in the spreadsheets that is the information you need to know to be predictor or you know, have a better forecast, which is like who may leave, what is the attrition that might come through, who's the low producers, where are we at with certain levels of the hiring process, like if it rather be a requisition on a spreadsheet, like what is the actual, like where are we in the process with recruiting, etc. So I think there's from a, I could go on for days about the structure of that data for use by finance teams or how it's used in things like Anaplan or Pigment, but it is a problem to be solved and there's a interaction between how you use that data with a Claude or any model that I think that's my whole thesis. There's an opportunity there for finance teams. But I'm curious to know what's the main feature you're using AI for with headcount? Like, what is the information that you're like, alright let me plug this into Claude, or let me plug this into some type of model. Here's the headcount data. This is the output I expect. What are you doing with that data today, and why is that important to you?
Reed Gilbert: So, it's the type of question that makes me think I'm not being creative enough or doing enough. Honestly, from a finance perspective, human costs are the vast majority of our spend. And so you have to have that stuff dialed. And there are probably more granular analyses I could do on benefit loading and taxes to better forecast or predict employee cost changes. But we're a remote company and that all goes out the window depending on which random state the next hire comes from and what their benefit situation is too. I have the luxury at Reprise of a pretty static and straightforward headcount model in cost. And we have pretty good regrettable attrition. So we don't spend a ton of time working on back fills or working on trying to figure out where the next back fill is coming from because it's pretty rare.
Eric Guidice: Sorry, I get it. I think the value of the tools is better for bigger data sets. Where, for example, we have a team that has thousands of delivery teams. This could be clinicians for a health care company, or security guards, or operators. And there's any hourly position has a normal life cycle that once you hit a certain level of tenure and whether you have room to grow in the compensation band, you could start to get predictive algorithmically about exits and then you could start to backfill people, especially if they're revenue producers, before the rolls actually open so that you decrease the ramp time of a revenue producer if you have anticipated attrition. So in bigger numbers when you have like hundreds or thousands of folks, that's where, you know, it really starts to return. But I wouldn't expect outside of the burden of employment calculation to have too much of a material impact on the financials of a 20 person firm. But I am very curious to know about if you're willing to share, right? I know that we're again getting into the more sensitive areas, right? But as you look at the dollars raised, are you raising less because of AI? Are you diluting less because of AI? Are you changing the structure of your Series B or your future financing strategy because of this productivity measure and what are your considerations as you balance headcount needs and fundraising?
Reed Gilbert: I'll say a couple of very general things about this. Mostly because fundraising and your capitalization strategy depends on many things, not just your team size and productivity. We expect to be a much smaller company for much longer than we had previously anticipated. And that increases your ARR per employee. It lets you distribute the same sized option pool amongst a smaller team and give everybody a more meaningful equity outcome. And these are the types of things that you love to do. Like it's, it's part of being in a tight knit team where everyone's rowing in the same direction and the glow of one team's wind carry right over to the other team. AI isn't for us and our business, it's not accelerating when we need to raise because of our AI spend. It's not decelerating when we need to raise because we have fewer people and we think we'll have more runway. Although, if anything, that may prove to be true and things are changing so rapidly that what we think now could be differently in two months time. But the biggest impact it's having is on our productivity and what we can achieve within a certain space of time. Because if you think about fundraising and that allowing you to get a multi-year period of growth, the features, the customers that we can reach all of the things that we do, can now do more of in the same amount of time. And so it's really empowering from the operator's side versus the investor's side.
Eric Guidice: Is there like a, when you're doing your board updates or preparing fundraising decks, is there a relationship metric between headcount and AI spend? Are you presenting any of this information in a different way from the previous startups that you were a leader within?
Reed Gilbert: Yeah, and the magnitude of the spend has changed over the past three months too. So, like, what was a line item before is now its own section. There are many reasons and things that could cause us to blow our budget and our plan for the year. AI spend is definitely one of those. And it's also probably the best way to miss your budget too. Sorry, we were being more productive than we thought we'd be. It was a better message to the board than we didn't forecast enough Salesforce seats.
Eric Guidice: I got it. Is there a relationship between AI spend and growth that like, yes, sorry, we, sorry, we were more productive. Do you also have to report on the, is it like a revenue coefficient that you're associating with that spend? You blow the budget by XYZ, do you have to have a proportionate or disproportionate return from a revenue or a growth side that justifies that spend or is all of that spend in this growth or build phase, you know, chalked up as R and D and experimentation and helping the company establish itself in its current trajectory.
Reed Gilbert: There's a lag time between when you ship the features and when you see the revenue flow through. And so we're really focused on adoption metrics for the things that we've been building recently. One of the byproducts has been that we're doing more demos and board sessions because our tech and our software is changing so rapidly and the board needs more frequent touch points and exposure to it to really grasp what's going on. Because it's one thing to say, hey, we built an MCP and now Claude can use our technology to make a demo. It's another thing to watch an agent make a demo with the transcript of the call you had this morning with a real customer and say, hey, make the demo that I want to show them in the follow-up meeting and use information from their website and this transcript and tailor it to these two people based on the questions they were asking today. It's like these are things that the best person could have done before with a significant amount of time spent that now an agent can do in the background. And keeping our board up to date on, hey, this is what's happening on the ground. This is why we're so excited right now. It takes more than metrics.
Eric Guidice: Who's the next finance, what's the first finance hire you make? What would you, if you had to hire somebody, what do you forecast being the first finance hire that you make? If at all. Like at how many employees, what do they do? What's the finance person you hire?
Reed Gilbert: That is past my planning horizon. As much as I love working with great people, the last place I want to put dollars to work is administration. And it's not to say that finance is all administrative, but with the leverage that you can get out of these tools and the way, probably more importantly, that these tools empower teams to do their own analysis and take ownership over things that they would have asked an FP and A analyst to do in the past. That's really cool.
Eric Guidice: I love it. Chris, what do you got? I know we're coming up at time. It's almost at the top of the hour.
Chris Mannion: Yeah. I was thinking of like one main takeaway. And it sounds like for our audience specifically, it sounds like if we could recommend that people spin up on code in Excel to be able to talk at the same language as their finance peers, I think that would be kind of a huge win. If that was kind of one thing we could recommend, is there anything, anything else for you to just over simplify what we spent 45 minutes talking about?
Reed Gilbert: I don't know if this is gonna run against the grain for everything you guys are doing, but just spend more time building and less time planning. Things are changing so fast, it's really hard to predict where we're gonna be in the next couple of quarters.
Eric Guidice: No, actually, it kind of reinforces what we're doing. Like, we're not trying to keep an old system alive. We're trying to get the data from the headcount process into the AI era. So I think if you've watched any of Chris's lessons on using Claude for HR or how to extract the information from a headcount data set using some of these tools, what you'll find, I mean, even when I demo some of the largest companies, there's still the data that you would try to get into an AI model is very unstructured, it's very difficult to have an agent use this information. So whether Chris's lessons within Claude are helping you organize it on your own or you're using a tool like headcount 365 to like plug into the HR recruiting and finance system. To unify the corporate taxonomy and have one ID for an individual role or user so that you can then have an agent understand the information. You know, that spend less time planning might be a great motto for headcount. Yeah, our current one is smarter workforce planning, but spend less time planning is a good, you might actually see on the website very shortly because that's, I think what you're getting at is what we're trying to do with this error like... Sales teams have Salesforce, sales teams have Gong. They have all these different ways to extract information from a process that an AI agent can then go use to create a demo. But if you ask a company to plug some of their unstructured data into a cloud model, it still needs work. So whether you're using Chris's lessons to organize it or a tool like mine, that's, I think, our general goal is to try to get people doing that. So not against the grain at all.
Chris Mannion: No, I think it reinforces my takeaway here is we were talking about OODA loops a while back and that comes from the military Observe, Orient, Decide, Act. And one of the things that we were saying there, our hypothesis was that OODA loops are getting tighter as the macro climate changes with AI and with competitive insight. So the faster you can move through that cycle, the more competitive you will be as an organization. I think everything that Reed said kind of reinforces that. You can plan in two minutes, which means you can move on to executing and figuring out what this means and then get feedback. And the faster feedback loop is going to empower the people that can do that to be more competitive in the market. And so I think it supports everything we've been saying, just probably a really interesting lens as we're coming in from the HR perspective. And I think it's good to have that finance viewpoint too.
Eric Guidice: Yeah, I love it. Reed, this was an awesome conversation. I really appreciate you spending, you know, with the AI tools you have, I'm not sure how much work we got in the way of by taking an hour of your time to talk about it, but I think our audience will appreciate it. I think it's a super interesting perspective about how AI is kind of disrupting the space and how you think about it as a finance professional. So, you know, thank you for spending time with us. I think we're going to have a few conversations like this over the course of the next couple months with different startups. CFOs, heads of talent, HR planners, some external vendors. So stay tuned for more headcount experts and look forward to recording the next episode.
The Finance & Headcount Relationship
In Episode 17 of the Headcount Experts podcast, hosts Chris Mannion and Eric Guidice sit down with Reed Gilbert, CFO at Reprise, to discuss the radical shift occurring at the intersection of finance and human resources. As AI tools evolve from "extracurricular" novelties into core drivers of productivity, the traditional 12-month headcount plan is being dismantled in favor of agile, token-based execution. This episode explores the transition from hierarchical "pyramid" organizations to leaner, AI-augmented structures where the focus shifts from managing people to managing information and output.
Key Takeaways from the Episode
1. The Death of the 12-Month Hiring Horizon
Traditional headcount decisions often operate on a year-long horizon: time to hire, time to ramp, and time to see results. However, with the current rate of AI advancement, a role defined today may be automated in three months. Gilbert argues for extreme discipline, suggesting that companies should only hire for roles that are absolutely essential to scaling or those that would be the "last person standing" if the company were to contract.
2. Trading Salary for Tokens
A significant shift is occurring in financial modeling. CFOs are increasingly viewing AI spend as a category of its own—one they are happy to "overspend" on. Unlike traditional software, AI usage (token consumption) is directly tied to immediate productivity gains. In many cases, increasing the AI budget per person is more cost-effective than adding new headcount, as it avoids the "orchestration overhead" of managing larger teams.
3. The "Circle" Org Chart
The organizational structure is moving from a pyramid to a circle. In this model, AI agents sit at the center, handling multi-disciplinary tasks, while human employees act as specialized managers of these agents. This allows for unconventional productivity, such as designers contributing significant amounts of code or product managers handling DevOps, effectively blurring the lines between traditional job descriptions.
4. Automating the Administrative Burden
Finance and HR roles are being transformed by AI fluency. Gilbert shares how complex, time-consuming tasks like quarterly board meeting option pool roll-forwards—which previously took hours of manual spreadsheet work—can now be completed in minutes using AI. This leverage allows startups to remain small and efficient for longer, delaying the need for administrative hires and increasing the equity value for the existing core team.
Headcount Finance is Changing: Build More, Plan Less
The overarching message for finance and talent leaders is clear: the competitive advantage now belongs to those with the tightest "OODA loops" (Observe, Orient, Decide, Act). In an era where AI models can change the capability of your workforce overnight, spending months on static workforce planning is a liability.
To thrive, organizations must prioritize clean, structured data that AI can interpret and focus on "building" rather than "planning." By embracing AI fluency and shifting the corporate mindset toward individual leverage, leaders can build more resilient, high-output organizations that achieve more with fewer, more empowered people.